Re: adding mod_reqtimeout to trunk?
On Monday 05 October 2009, Jim Jagielski wrote: Thx... I'm updating it with an eye to making it core, and therefore having ReqTimeout headerinit=5 headermax=10 As we also have RequestHeaders, maybe RequestTimeout would be better? Let me know if I can help w/ the docs. I have commited some docs. Any improvements are welcome. BTW, how will the timeouts interact with SSL renegotiation and with mod_request? I haven't tried that yet. Cheers, Stefan
Re: adding mod_reqtimeout to trunk?
Thx... I'm updating it with an eye to making it core, and therefore having ReqTimeout headerinit=5 headermax=10 Let me know if I can help w/ the docs. On Oct 4, 2009, at 3:40 PM, Stefan Fritsch wrote: On Sunday 04 October 2009, Nick Kew wrote: FWIW, IMO it should go in modules/filters not experimental. +1. trunk is, by definition, experimental. But when we float off 2.3/4-branch, we should perhaps do some documentation of stability levels of different features and modules for users. I might open a wiki page to collect information on the subject. I agree and would rather reserve modules/experimental for modules that have known issues. I have commited mod_reqtimeout to trunk. I haven't finished the docs, yet.
Re: adding mod_reqtimeout to trunk?
On 10/04/2009 10:23 AM, Stefan Fritsch wrote: Hi, I would like to add mod_reqtimeout [1,2] to trunk. Is this OK? Considering the positive comments it received, may I put it into modules/filter or should it go into modules/experimental first? I guess experimental is a good starting point. Regards Rüdiger
Re: adding mod_reqtimeout to trunk?
+1 Rainer Jung wrote: On 04.10.2009 10:23, Stefan Fritsch wrote: Hi, I would like to add mod_reqtimeout [1,2] to trunk. Is this OK? I think it would be a useful addition. Regards, Rainer
Re: adding mod_reqtimeout to trunk?
On 4 Oct 2009, at 09:23, Stefan Fritsch wrote: Hi, I would like to add mod_reqtimeout [1,2] to trunk. Is this OK? Considering the positive comments it received, may I put it into modules/filter or should it go into modules/experimental first? experimental has been somewhat in limbo of late: there was some discussion on here that led to it no longer being used. I don't recollect why, but if you use it for mod_reqtimeout, there are quite a number of other new modules that should join it there: in fact probably most of what's new in trunk over 2.2. -- Nick Ke
Re: adding mod_reqtimeout to trunk?
Personally, I'd like to see this as part of the actual code core, where we have several Timeouts, eg: Timeout 30 5 10 2 which define timeout as now, timeout before 1st byte, timeout between bytes timeout after etc... We've always wanted better control over this ind timeouts and putting it in a module seems not a good idea for 2.4/3.0 (but of course, OK for 2.2.x) On Oct 4, 2009, at 4:23 AM, Stefan Fritsch wrote: Hi, I would like to add mod_reqtimeout [1,2] to trunk. Is this OK? Considering the positive comments it received, may I put it into modules/filter or should it go into modules/experimental first? Cheers, Stefan [1] http://www.sfritsch.de/mod_reqtimeout/mod_reqtimeout.c [2] http://mail-archives.apache.org/mod_mbox/httpd- dev/200908.mbox/200909010043.54040...@sfritsch.de
Re: adding mod_reqtimeout to trunk?
On Sunday 04 October 2009, Jim Jagielski wrote: Personally, I'd like to see this as part of the actual code core, where we have several Timeouts, eg: Timeout 30 5 10 2 which define timeout as now, timeout before 1st byte, timeout between bytes timeout after etc... We've always wanted better control over this ind timeouts and putting it in a module seems not a good idea for 2.4/3.0 (but of course, OK for 2.2.x) I think putting mod_reqtimeout into trunk until it is backported to 2.2 and then moving it into the core would make sense, wouldn't it? And I would prefer several config directives instead of having to remember which value in Timeout means what.
Re: adding mod_reqtimeout to trunk?
On Oct 4, 2009, at 2:38 PM, Stefan Fritsch wrote: On Sunday 04 October 2009, Jim Jagielski wrote: Personally, I'd like to see this as part of the actual code core, where we have several Timeouts, eg: Timeout 30 5 10 2 which define timeout as now, timeout before 1st byte, timeout between bytes timeout after etc... We've always wanted better control over this ind timeouts and putting it in a module seems not a good idea for 2.4/3.0 (but of course, OK for 2.2.x) I think putting mod_reqtimeout into trunk until it is backported to 2.2 and then moving it into the core would make sense, wouldn't it? +1 And I would prefer several config directives instead of having to remember which value in Timeout means what. Well, I'm not a big fan of directive creep, but I see your point and agree with it in a general sense. FWIW, IMO it should go in modules/filters not experimental.
Re: adding mod_reqtimeout to trunk?
Jim Jagielski wrote: And I would prefer several config directives instead of having to remember which value in Timeout means what. Well, I'm not a big fan of directive creep, but I see your point and agree with it in a general sense. Timeout Thistimeout=10 Thattimeout=2 Othertimeout=300 ? We have precedent for this kind of halfway house. FWIW, IMO it should go in modules/filters not experimental. +1. /trunk/ is, by definition, experimental. But when we float off 2.3/4-branch, we should perhaps do some documentation of stability levels of different features and modules for users. I might open a wiki page to collect information on the subject. -- Nick Kew
Re: adding mod_reqtimeout to trunk?
Jim Jagielski wrote: On Oct 4, 2009, at 2:38 PM, Stefan Fritsch wrote: On Sunday 04 October 2009, Jim Jagielski wrote: Personally, I'd like to see this as part of the actual code core, where we have several Timeouts, eg: Timeout 30 5 10 2 which define timeout as now, timeout before 1st byte, timeout between bytes timeout after etc... And I would prefer several config directives instead of having to remember which value in Timeout means what. Well, I'm not a big fan of directive creep, but I see your point and agree with it in a general sense. I agree with Stefan, 30 5 10 2 is nonsensical to the casual administrator. But going to your point, Jim, perhaps using some sort of Timeout keepalive=5 syntax might be good? Directives are in a hash, and only evaluated for .htaccess at runtime, so it's not the place to focus our optimization energies.
Re: adding mod_reqtimeout to trunk?
On Sunday 04 October 2009, Nick Kew wrote: FWIW, IMO it should go in modules/filters not experimental. +1. trunk is, by definition, experimental. But when we float off 2.3/4-branch, we should perhaps do some documentation of stability levels of different features and modules for users. I might open a wiki page to collect information on the subject. I agree and would rather reserve modules/experimental for modules that have known issues. I have commited mod_reqtimeout to trunk. I haven't finished the docs, yet.
